>> * If a task requests CLONE_PPID but does not have PF_PPIDOK - should we
>> fail clone() or silently ignore CLONE_PPID
> I'd go for failing clone. You'd want clone to fail if it couldn't give
> you the new vm you wanted, for example.
It's better then to do not what program asked. But what about combining
efforts ?
Now we have:
CLONE_PPID/CLONE_PPIDOK/PF_PPIDOK by Tim Hockin <thockin@isunix.it.ilstu.edu>
CLONE_SUSPENDED by Ulrich Drepper <drepper@cygnus.com>
newpgid/getpgids/joinpgid/leavepgid/sigsemantics by
Larry McVoy <lm@bitmover.com> and Chris Smith <cd_smith@ou.edu>
(not yet done) ...
CLONE_SUSPEND is usable even without threads (usually Linux's behaviour
where child is activated after fork and parent is suspended is better then
BSD behaviour but sometimes it's not true and why not add such ability if
it's just 6 lines of code in kernel).
Looks like all three things must be added to kernel for proper support of
POSIX threads. And LinuxThreads must be patched and TESTED. And only then
ONE letter to Linus must be created. If Linus will get 3-5-10 unrelated
conflicting patches for better threads support he'll just reject them all...
But if it'll be only one patch with clear explanation and approval of Larry,
Drepper (and may be Alan Cox as well) then this patch can be accepted...
Why are you so interested in scattering efforts ??
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/